Setting up a Machine Vision System for Robotics-Based Manufacturing

Machine vision is a collective term representing the image processing technology and cameras that are used in robotics manufacturing systems. Machine vision is a manufacturing approach that offers flexibility to ensure that robots locate parts and operate intelligently without requiring mechanical or human intervention. In machine vision powered systems, robots utilize digital image processing and cameras to recognize products and locate them on a conveyor that exits from an injection moulding machine. A good example is when a machine vision system triggers a robot to pick a product and assemble it or manipulate it to apply a label.

There are numerous things that should be considered when setting up a machine vision robotic manufacturing system, including these four steps.

Determining camera resolution and lighting requirements

The ideal camera resolution is determined by the field of view or area the camera must focus on and how accurately the robot should be positioned in order to locate required features. As such, large fields of view that require precise positioning must use a high camera resolution. However, higher camera resolutions require longer image processing durations and expensive hardware.

Most machine visions systems require an effective and dedicated source of lighting that minimizes image variations and accentuates product features like unique marks or edges that the camera is trying to capture. Therefore, the importance of having an effective lighting system cannot be understated.

Configuring the system

Robot control software, image processing software, the user interface and communication between the camera, robot and host PC must be configured accordingly. The configuration process can be made easier by using a development framework that supports different hardware devices and application tools.

Calibrating the system

The machine vision system must be calibrated in order to provide accurate results. System calibration entails camera calibration, camera-to-robot coordinate calibration, compensation for variations in product heights and setting up the part locator.

The camera is calibrated to remove radical distortion and perspective in images that may be caused by the lens properties and camera orientation. Camera calibration provides useful measurements in real units like millimeters as opposed to pixels. Camera-to-robot coordinate calibration ensures that the robot uses coordinates determined by the camera. Compensation for variations in product height ensures that errors are avoided by performing checkerboard calibration while the checkerboard’s height from the conveyor surface is similar to that of the part where image processing extracts features to locate the required part. Setting up the part locator entails using different techniques like line finders, pattern locators or blob tools to locate features on a specific part.

Operating and maintaining a machine vision system

A machine vision system does not only require software configuration, hardware installation and system testing but also good operation and maintenance. This is because the system may be required to manufacture new products. Therefore, it must have an easy to use user interface to ensure faster and efficient product feature changes and system calibrations. Software-based machine vision can offer significant advantages toward these objectives.

Matt Edwards is the Director of Professional Services for KINGSTAR an IntervalZero Company.